Compartés, holiday 2025: a bigger season inside a 4x floor
Winner, 2026 Meta Agency Awards: Performance Campaign of the Year by a Boutique Agency, North America. Campaign dates: November 1 to December 31, 2025. My role: I helped lead the elk team, built the Meta budget plan, set the weekly budget steps, directed creative needs and wrote the award submission.
Compartés sells premium chocolate, and the holiday season decides its year. The job was to reach more gift buyers without spending the budget too early. We finished the season with $2.84 million in Shopify sales, up 33%. Over Black Friday weekend we nearly doubled Meta spend and still lowered the cost of a purchase.
This page walks through the plan, the account structure, the ads and the results. The source documents are public too: the planning excerpts, the full award submission and a PDF evidence pack.
My part and the team's
I created the two planning tasks that ran the season. The campaign update task, written November 5 and last edited November 21, set the pace and the creative asks. Its first instruction:
Increase budgets 10-20% WoW
It also asked for more compilation videos, holiday creator content, photography, stronger promotion of the advent calendar and a daily check of spend, revenue and return. The budget-plan task, created November 7, holds the $200,000 Meta plan below.
Creators, designers and other strategists made the ads and ran the account day to day. The award recognizes that shared work. I used AI for copy drafts, research and analysis along the way. Every call on budget and creative was made by a person.
The target
Premium chocolate is a crowded holiday category. Buyers compare prices and worry about delivery dates. Compartés has unusual flavors and striking packaging, so the ads had to make those easy to see.
We agreed on an account-level Meta return floor of 4x: at least four dollars of attributed purchase value per dollar of spend. We read it alongside store sales and total ad cost, because return on ad spend measures revenue, not profit.
The $200,000 plan
The plan split the season by how people shop. Early November was light testing. The two weeks before Black Friday built momentum. The sale and the week after took the most money. December followed gifting and delivery cutoffs, then post-holiday self-gifting.
Scroll sideways to see all columns.
| Stage | Dates | Planned Meta budget | Purpose |
|---|---|---|---|
| Early warm-up | November 1–7 | $15,000 | Light testing while demand is low |
| Before Black Friday | November 8–21 | $40,000 | Build toward the sale and test creative |
| Sale and extended peak | November 22–December 8 | $85,000 | Most money into the strongest demand |
| Holiday gifting | December 9–20 | $45,000 | Gift purchases before delivery cutoffs |
| After the holiday | December 21–31 | $15,000 | Lower competition and self-gifting |
| Total | November–December | $200,000 | Five stages, adjusted daily |
Final Meta spend was $201,210.57, within 1% of the plan. Budgets went up 10 to 20% at a time, and only after the previous step held.
Four campaigns, four jobs
Each campaign had one commercial job, so a result always pointed at a decision.
Scroll sideways to see all columns.
| Campaign | Job | Spend | Meta return |
|---|---|---|---|
| Evergreen | Keep proven ads selling | $121,805.49 | 5.29x |
| Black Friday promotion | Give buyers a reason to act on the offer | $25,514.00 | 4.42x |
| Shopping and catalog | Show the right products from the catalog | $22,820.86 | 3.44x |
| Seasonal products | Launch limited editions and the advent calendar | $29,064.53 | 3.17x |
About 87% of consumer spend used Meta's Advantage+ Audience with open placements, guided by Shopify audiences, customer segments, purchaser exclusions and selected lookalikes. Separate tests challenged that default: a household-income and online-shopping test returned 7.12x.
The ads
These are the original campaign assets from the award submission, with the spend, return and click-through rate reported for each. Videos and the GIF appear as stills.
Creator video
Spend $13,955.15, return 6.15x, click-through 0.93%. The highest-spend featured ad. Product in someone's hands makes the size, variety and gift appeal obvious.
Pistachio creative
Spend $11,277.32, return 7.12x, click-through 0.76%. First ran in June 2024 and was still selling in the 2025 holiday refresh. New ideas went in beside proven ones, not in place of them.
Evergreen compilation
Spend $10,818.38, return 5.38x, click-through 1.13%. Several creators in one video, so the campaign never rested on one face or one message.
Black Friday compilation
Spend $9,466.06, return 4.65x, click-through 1.82%. The format my update task asked for, applied to the sale.
Sale static
Spend $8,220.54, return 3.83x, click-through 1.48%. The offer is readable at a glance. It returned less than the evergreen winners, which is normal for a promotion ad.
The featured catalog ad, assembled by Meta from the product feed, spent $9,467.88 at 3.35x with a 2.19% click-through rate. Holiday creator round two returned 5.79x against 5.02x for round one. These are observed results, not controlled tests.
Black Friday: nearly double the spend, cheaper purchases
Same weekdays each year: November 24 to December 2, 2025, against November 25 to December 3, 2024.
Scroll sideways to see all columns.
| Meta measure | 2024 | 2025 | Change |
|---|---|---|---|
| Ad spend | $33,032 | $65,726 | +99.0% |
| Attributed purchase value | $135,422 | $277,208 | +104.7% |
| Attributed purchases | 1,012 | 2,233 | +120.7% |
| Return on ad spend | 4.10x | 4.22x | +2.9% |
| Cost per purchase | $32.64 | $29.43 | −9.8% |
| Cost per 1,000 impressions | $33.31 | $22.67 | −31.9% |
| Click-through rate | 1.395% | 1.508% | +8.1% relative |
Spend doubled and purchase value grew slightly faster. That was the evidence we needed to keep the peak budget high.
The full season: more volume, a planned dip in return
November 1 to December 31 in each year.
Scroll sideways to see all columns.
| Meta measure | Holiday 2024 | Holiday 2025 | Change |
|---|---|---|---|
| Ad spend | $125,927 | $201,211 | +59.8% |
| Attributed purchases | 4,646 | 6,924 | +49.0% |
| Attributed purchase value | $636,562 | $926,640 | +45.6% |
| Click-through rate | 0.980% | 1.375% | +40.3% relative |
| Cost per click | $1.55 | $1.28 | −17.1% |
| Adds to cart | 34,410 | 78,509 | +128.2% |
| Cost per add to cart | $3.66 | $2.56 | −30.0% |
| Checkouts started | 9,807 | 18,860 | +92.3% |
Full-season Meta return works out to about 5.06x in 2024 and 4.61x in 2025. Return fell while purchases and sales grew, and it stayed above the 4x floor. That was the trade we planned: more of the business, inside an agreed efficiency limit.
Store results
Shopify shows the whole store, all channels included.
Scroll sideways to see all columns.
| Shopify measure | Holiday 2024 | Holiday 2025 | Change |
|---|---|---|---|
| Orders | 15,982 | 21,112 | +32.1% |
| Total sales | $2,136,002 | $2,843,922 | +33.1% |
| Average order value | $126.23 | $126.93 | +0.6% |
| First-time customer orders | 8,303 | 11,800 | +42.1% |
| Returning-customer orders | 7,679 | 9,312 | +21.3% |
Growth came from more orders, mostly from new customers, not from bigger baskets. Combined Meta and Google spend rose 19.2% against 33.1% store growth, so Shopify sales per ad dollar went from 7.40x to 8.26x.
Scroll sideways to see all columns.
| Ad spend | Holiday 2024 | Holiday 2025 | Change |
|---|---|---|---|
| Meta | $125,926.72 | $201,210.57 | +59.8% |
| Google Ads | $162,764.26 | $142,957.99 | −12.2% |
| Combined | $288,690.98 | $344,168.56 | +19.2% |
Money moved toward Meta. Google reported 28.8% more conversions on less spend; the two platforms count differently, so those are not added together.
How the numbers were checked
Both years use Meta's seven-day click, one-day view attribution. Meta measured delivery and attributed purchases, Shopify measured store sales and customer mix, and Triple Whale cross-checked the reporting. Pixel and server-side purchase events supported measurement, with an Event Match Quality score of 9.3.
Read the full award submission · Read the original planning excerpts · Download the PDF evidence pack